Br J Ophthalmol: first published as 10.1136/bjo.65.5.323 on 1 May 1981. Downloaded from

British Journal of Ophthalmology, 1981, 65, 323-328 Uveitis, vitreous humour, and klebsiella. II. Cross-reactivity studies with

J. WELSH, H. AVAKIAN, AND A. EBRINGER From the Udfit, Department of Biochemistry, Queen Elizabeth College, London W8, and the Department of Rheumatology, Middlesex Hospital, London WI

SUMMARY Radioimmunoassay with calf and cow vitreous humour-1251 and rabbit antivitreous humour serum has been employed to investigate the immunological cross-reactivity of vitreous humour with bacterial and mammalian tissue antigens. Klebsiella ultrasonicate preparation at a dose of 10 000 ,ug/ml was found to inhibit the binding of vitreous humour by 25-100% (p <0-0001), compared with an inhibition of 5-30% by a similar quantity of E. coli ultrasonicate preparation. Equivalent amounts ofStreptococcus pyogenes antigen, bovine haemoglobin, and hyaluronic acid had no inhibitory effect, while horse spleen ferritin was found to inhibit vitreous humour binding between 0 and 10%. These results indicate that klebsiella micro-organisms have antigens which partially resemble some eyeball components. It is suggested that acute anterior uveitis of ankylosing spondylitis may be produced by anti-Gram-negative bacterial binding to cross-reacting eye antigens.

Acute anterior uveitis (AAU) or iridocyclitis is an Freund's adjuvant several times over a period of 6 inflammatory condition of the uveal tract which may months as described in the previous paper." After 9 occur in association with a number of arthritic immunisations the rabbit was bled, the serum was disorders including ankylosing spondylitis (AS), separated and precipitated with half saturated Reiter's syndrome, and Still's disease. The reason for ammonium sulphate, and dialysed against phosphate this association is unknown. However, an increased buffered saline (PBS), and the resulting antibovine

frequency of the histocompatibility allele HLA B27 vitreous humour immunoglobulin (anti-VH-Ig) http://bjo.bmj.com/ has been found in these diseases.'-3 Klebsiella preparation (Al) was stored at -15°C until use. A2 pneumoniae has been isolated from faecal cultures of immunoglobulin preparation was obtained after 11 patients with active AS4 as well as from AS patients immunisations. Preparations were restored to with uveitis.'8 K. pneumoniae has been found to original volume with PBS. partially cross-react with antigenic determinants located on HLA B27 lymphocytes,7 8 and some of PREPARATION OF BACTERIAL ANTIGENS these could involve the HLA B27 antigen itself,9 or Klebsiella pneumoniae (MXlO0) was isolated from a a closely associated gene product.10 faecal culture of an ankylosing spondylitis patient on September 24, 2021 by guest. Protected copyright. In an endeavour to determine a possible patho- and the bacterial extract 'klebsiella sonicate prepara- logical connection between Gram-negative micro- tion' (KSP) was obtained as previously described.9 organisms and uveitis the structural properties of Briefly, klebsiella micro-organisms were grown on klebsiella miro-organisms have been investigated by MacConkey culture plates, harvested, washed twice competitive inhibition in a vitreous humour radio- with cold PBS (4°C), resuspended (20 ml; 1-3 x 1013 with bacterial and mammalian tissue micro-organisms), ultrasonicated, and centrifuged. antigen extracts. Supernatant was filtered through Sephadex G25, dialysed against distilled water, and lyophilised. Materials and methods 100 mg residues were recovered which dissolved readily in PBS and EI%m at 280 nm was 11-8. A similar PREPARATION OF VITREOUS HUMOUR bacterial extract was ANTISERUM prepared from Escherichia coli A New Zealand white rabbit was immunised with (ML 30) micro-organisms, and streptococcus group crude cow-vitreous-humour (CVH) in incomplete A antigen was obtained commercially (Difco).

Correspondence to Dr A. Ebringer, Immunology Unit MAMMALIAN TISSUE ANTIGENS Department of Biochemistry, Queen Elizabeth College, Mammalian tissue antigens used as controls in the London W8 7AH. comparative studies were (i) hyaluronic acid prepared 323 Br J Ophthalmol: first published as 10.1136/bjo.65.5.323 on 1 May 1981. Downloaded from

324 J. Welsh, H. Avakian, and A. Ebringer from human umbilical cord (British Drug Houses); out in quadruplicate and percentage bound (%B) (ii) haemoglobin obtained from beef blood (Sigma); was expressed as: (iii) ferritin extracted from horse spleen (Miles Laboratories). Percentage bound= /g= Test(cpm)-Blank(cpm) x 100. [ Total(cpm) - Blank(cpm) J

PREPARATION OF VITREOUS HUMOUR Bovine vitreous humour was homogenised in a SELECTION OF ANTISERUM DILUTION FOR Potter-Elvejhem homogeniser and centrifuged to RADIOIMMUNOASSAY remove any large particulate matter. The vitreous The dilution of antiserum that binds a given percen- humour was then ultrasonicated in an MSE 25 tage of labelled antigen for use in the radioimmuno- ultrasonicator, at medium setting, with 10-second assay (RIA) was selected from binding curves, bursts for 4 cycles, recentrifuged, and supernatant described in the previous paper,"' and was taken as dialysed continuously against 10 litres of deionised the dilution giving maximum difference in percentage water before being lyophilised. Calf vitreous humour bound of CVH-1251, when comparing antivitreous obtained from younger animals was treated in a humour serum with preimmunisation serum from the similar manner, but, after ultrasonication, the CVH same rabbit. For instance, for a particular batch of was centrifuged in an MSE 18 centrifuge for 1 hour vitreous humour, at a dilution of 1/2000 of anti-VH- at 12 000 rpm at 4°C and then dialysed overnight at Ig there was 30% binding of CVH-1251 compared 4°C against 10 litres of PBS containing 0 1 % sodium with less than 1 % binding by preimmunisation azide (PBS-A). The CVH was then desalted on fine serum immunoglobulin, and therefore in this Sephadex G25 (Pharmacia) before being lyophilised. particular assay system percentage bound in absence of competitor (BO) equals 30% (Fig. 1). For each new batch of labelled vitreous humour a radio- IODINATION AND FRACTIONATION OF binding assay was carried out to determine the VITREOUS HUMOUR dilution point of maximum difference in binding Crude bovine vitreous humour (CVH) (2 mg) was between pre- and postimmunisation sera prepara- radiolabelled with 1 mCi of Na1251 (Radiochemical tions for use in RIA. Centre, Amersham) as described in the previous paper.'1 After dialysis the 1251-labelled vitreous RADIOIMMUNOASSAY humour was applied to a Sepharose 6B column In the radioimmunoassay the concentrations of (Pharmacia), equilibrated with 0-15 M PBS, pH 7-2, and radiolabelled tracer remain constant, containing sodium azide (PBS-A), and the first and increasing amounts of competitor are added to http://bjo.bmj.com/ fraction was collected as described," stored at 4°C, displace the label. A standard competition or and used in subsequent assays. inhibition curve is obtained with unlabelled CVH and compared with inhibition produced by equal RADIOBINDING ASSAY quantities of other antigens. To 50,ul aliquots of The assay buffer used in all experiments was 0-15 M CVH-1251, diluted as for radiobinding assay, was. PBS-A containing 04% bovine serum albumin added 50 ,tl of buffer containing increasing concen- (BSA) (Radioimmunoassay grade, Sigma). To 50 ,ul trations of competitor, then 50 ,ul of anti-VH-Ig at on September 24, 2021 by guest. Protected copyright. assay buffer in an LP3 tube (Luckhams) was added the dilution determined by the radiobinding assay 50 l of tracer CVH-1251 followed by 50 l of rabbit and incubated at 37°C for 30 minutes. After cooling, anti-VH-Ig in serial dilutions. After 30 minutes' 50VOl of coprecipitating serum was added, and the incubation at 37°C, 50 ,ul of coprecipitating donkey tubes were left overnight at 4°C. The washing antirabbit immunoglobulin G serum (Wellcome procedure was carried out as for the radiobinding Reagents) was added. The optimum dilution of assay, radioactivity counted in a gamma counter, second antibody was determined by measuring the and residual binding with (%B) and without com- quantity of first antibody precipitated by varying petitor (%Bo) determined as before. Percentage dilutions of coprecipitating antibody. Maximum inhibition (%I) was calculated as follows: precipitation was obtained with a dilution of 1/180 of coprecipitating donkey antirabbit IgG serum per Percentage inhibition= %I= ( 1-B- x 100. 10 ,tg of anti-VH-Ig. The tubes were then left over- BoZ night at 4°C, washed twice with PBS-A containing Results 0-01 M potassium iodide, centrifuged at 3000 rpm for 30 minutes, and the radioactivity in the precipi- CALF VITREOUS HUMOUR tate counted in a well-type gamma counter (Packard RADIOIMMUNOASSAY Model 578). Each serum dilution assay was carried The antiserum dilutions showing maximal binding Br J Ophthalmol: first published as 10.1136/bjo.65.5.323 on 1 May 1981. Downloaded from

Uveitis, vitreous humour, and klebsiella 325

VITREOUS HUMOUR RADIOIMMUNOASSAY Vitreous Humour Bo0l30% Antiserum /2000

Fig. 1 Inhibition ofcalf CVH-1251 tracer binding to Inhibition rabbit anti- VH-Ig, at a dilution of 1/2000, by increasing amounts (lg/ml) ofunlabelled vitreous humour, klebsiella sonicate preparation (KSP), ferritin, bovine haemoglobin, and Kiebsiella S.P. hyaluronic acid. Bars represent means ± standard errors.

Ferritin

100 1000 10000 Competitor pg/m were selected for inhibition studies and were antigen had no inhibitory[effect. The difference in obtained from radiobinding curves prepared for each inhibition between KSP and ferritin, when Bo equals batch of radiolabelled vitreous humour. RIA 54% (Fig. 2) is statistically significant (t=20-56, studies, with 2 dilutions being used at 1/2000 p <0=001). Lower concentrations of competitor (Bo= 30 %) and 1/840 (Bo = 54 %) respectively, of A2 produced correspondingly lower levels of inhibition. preparations of rabbit anti-VH-Ig with calf-1251 are shown in Figs. I and 2. COW VITREOUS HUMOUR At 10 000 ,ug/ml of competitor, klebsiella extract RADIOIMMUNOASSAY (KSP) showed 26 % (Fig. 1) and 30 % (Fig. 2) RIA studies with rabbit anti-VH-Ig preparations, http://bjo.bmj.com/ inhibition of vitreous humour binding respectively, obtained from 2 separate bleeds (A2 and Al), while ferritin inhibited less than 10% and bovine binding to cow vitreous humour are shown in haemoglobin, hyaluronic acid, and streptococcus Figs. 3 and 4.

VITREOUS HUMOUR RADIOIMMUNOASSAY Vitreous Hlumour

80 on September 24, 2021 by guest. Protected copyright. B0-54% Antiserum 1 (840 70

60 Fig. 2 Inhibition of calf CVH-1211 tracer binding to 50 rabbit anti- VH-Ig, at a dilution of 1/840, by increasing amounts I nhibition ([ig/ml) ofunlabelled vitreous 40t humour, klebsiella sonicate Klebsiella S. P. preparation, ferritin, hyaluronic S*P. acid, and streptococcus A antigen. Bars represent means 201 standard errors.

Ferriti n 10 Br J Ophthalmol: first published as 10.1136/bjo.65.5.323 on 1 May 1981. Downloaded from

326 J. Welsh, H. Avakian, and A. Ebringer

VITREOUS HUMOUR RADIOIMMUNOASSAY

80

Klebsiella S.P. Fig. 3 Inhibition ofcow 60 CVH-1261 tracer binding to rabbit anti- VH-Ig (A2), at a I nhibition dilution of1/300, by increasing amounts (g/ml) ofunlabelled vitreous humour, klebsiella and 40 E. coli sonicate preparations, and ferritin. Bars represent means ± standard errors. 20

E. coli S. P. Ferritin 10 100 1000 10000 Competitor lj/m1 At 10 000 ,ug/ml of competitor, klebsiella extract (t=11P76, p<0001). An equal amount of E. coli (KSP) showed 65-5 % inhibition of vitreous humour sonicate preparation showed 28% inhibition and binding by anti-VH-Ig (A2) at a dilution of 1/300 ferritin 10%. Lower concentrations of competitor (Bo= 79 %), while an equal amount of E. coli extract again produced correspondingly smaller degrees of produced 4'8 % of inhibition, and ferritin had no inhibition. effect (Fig. 3). Another preparation of anti-VH-Ig (Al), at a Discussion dilution of 1/110 (BO=47%), showed 100% inhibi- http://bjo.bmj.com/ tion with 10 000 ,ug/ml of klebsiella sonicate prepara- Antigenic similarities between vitreous humour and tion (KSP), and this difference is statistically klebsiella micro-organisms were suggested by the significant when compared with all other competitors studies described in the previous paper, namely,

VITREOUS HUMOUR RADIOIMMUNOASSAY

Bo ' 47% 1 on September 24, 2021 by guest. Protected copyright. Antiserum /110

Fig. 4 Inhibition of cow CVH-1251 tracer binding to rabbit anti- VH-Ig (Al), at a dilution of 1/110, by increasing amounts (Vg/ml) ofunlabelled vitreous humour, klebsiella, and E. coli sonicate preparations and ferritin. Bars represent means ± standard errors. E. coli S. P.

Ferritin

10 100 1000 10000 Competitor lj/ Br J Ophthalmol: first published as 10.1136/bjo.65.5.323 on 1 May 1981. Downloaded from

Uveitis, vitreous humour, and klebsiella 327 binding studies with labelled vitreous humour. Klebsiella are non-motile, capsulate, Gram- Rabbit antisera raised against klebsiella were found negative bacilli of the family Enterobacteriaceae, to bind vitreous humour to a greater extent than frequently found in the lower gastrointestinal tract sera raised against other bacterial or viral antigens. of man. Investigations into the composition of Radioimmunoassay studies described here show capsular polysaccharides of klebsiella have demon- that klebsiella antigens directly inhibit the binding strated the presence of glucuronic acid and various of vitreous humour antigen to antivitreous humour monosaccharides such as mannose and galactose,"3 antibodies. A similar preparation from another and a similar polysaccharide composition has been Gram-negative micro-organism, Escherichia coli, described for vitreous humour.'4 Lipopolysaccharides showed a lower level of inhibition, while other extracted from Gram-negative bacteria have been mammalian antigens, such as ferritin, hyaluronic shown to specifically inhibit lymphocytotoxic acid, and bovine haemoglobin showed little or no activity of various HLA antisera,15 and it was inhibitory effect. A bacterial preparation from a suggested that lymphocytotoxins evoked by bacterial Gram-positive micro-organism, streptococcal group antigens could affect the host's own tissues. Thus A antigen, also had no inhibitory effect on the binding cross-reactive antibodies, produced after bacterial of vitreous humour antigen by vitreous humour anti- infection with micro-organisms such as klebsiella, serum. The klebsiella preparation has greater which bind to vitreous humour antigens, as sug- inhibitory effect when cow rather than calf vitreous gested by this study, could be involved in the humour is used in the radioimmunoassays. This development of anterior uveitis. The acquisition of lower level of inhibition may either represent an enteric flora has been suggested as a possible differences in vitreous humour composition between cause for the development of isohaemagglutinins in young and old animals or could be due to the anti- neonatal animals, including man,'6 and thus the serum raised in the rabbit, which had been immunised appearance of cross-reacting antibodies against with vitreous humour obtained from eyeballs of old other tissues, such as vitreous humour, could also cows. occur after infection of the gastrointestinal tract Cross-reactivity in radioimmunoassay systems with Gram-negative micro-organisms. depends on the Bo value, which arises from the Antibodies produced after an infection by relationship between bound and unbound fractions klebsiella could bind to tissues containing large of radiolabelled tracer and affinity of antibodies in quantities of cross-reacting material such as vitreous the antiserum used in the assay system. Immuno- humour and other eye antigens, activate the comple-

chemical studies using haptenic systems have shown ment cascade, and produce an inflammatory http://bjo.bmj.com/ that the higher affinity antibodies interact to a response. Acute anterior uveitis could be a manifesta- greater extent with antigen analogues than lower tion of such a process, possibly through the adjuvant affinity antibodies isolated from the same serum.'2 effect of some concurrent infection.'7 In experimental The relationship of the Bo value to cross-reactivity animal models, such as adjuvant arthritis, a systemic is therefore the result of a very complex dynamic disease is produced which is characterised by equilibrium between antibody affinity, amount of arthritis of peripheral joints and spinal column, tracer, and the affinity of the competitor for the together with a nongranulomatous uveitis.'8 It has antibody binding site. At present there are no been suggested that adjuvant-induced uveitis is on September 24, 2021 by guest. Protected copyright. satisfactory physicochemical models for multisite probably the result of an immunological response of antigen-antibody systems, while precise measure- the hypersensitivity type to bacterial antigens,1920 ments of affinity constants for these interactions are and the demonstration of cross-reactivity between lacking. Therefore, the assessment of cross-reactivity klebsiella micro-organisms and vitreous humour in in such systems requires caution, and comparisons this study is consistent with such a concept. can be made only between the different competitors. It is suggested that uveitis in patients with The data presented here consist of comparisons ankylosing spondylitis could be produced by anti- between the homologous antigen, vitreous humour, bodies evoked after gastrointestinal infection by and a number of heterologous antigens, each having Gram-negative bacteria. a varying degree of biochemical homogeneity and immunological heterogeneity. The preparation from We thank Dr A. C. Boyle and the Arthritis and Rheumatism Council for support. J. W. was in receipt of an SRC student- klebsiella micro-organisms was found to compete ship. for antivitreous humour antibodies, compared to other antigens of similar heterogeneity, and one References possible explanation is that klebsiella micro- 1 Brewerton DA. HLA-B27 and the inheritance of sus- organisms have antigens which resemble some ceptibility to rheumatic disease. Arthritis Rheum 1976; vitreous humour components. 19: 656-68. Br J Ophthalmol: first published as 10.1136/bjo.65.5.323 on 1 May 1981. Downloaded from

328 J. Welsh, H. Avakian, and A. Ebringer

2 Woodrow JC. HLA 27 and Reiter's syndrome. Lancet surface marker on the lymphocytes of patients with 1973; ii: 671-2. ankylosing spondylitis. Nature 1979; 277: 68-70. 3 Mapstone R, Woodrow JC. Acute anterior uveitis and 11 Avakian H, Abuknesha R, Welsh J, Ebringer A. Uveitis, HLA-B27. Lancet 1974; i: 681-2. vitreous humour, and klebsiella. I. Binding studies with 4 Ebringer RW, Cawdell DR, Cowling P, Ebringer A. rabbit antisera. Br J Ophthalmol 1981; 65: 315-22. Sequential studies in ankylosing spondylitis: association 12 Exley D, Avakian H. The relationship of specificity to of Klebsiella pneumoniae with active disease. Ann Rheum affinity of anti-hapten sera. J Steroid Biochem 1977; 8: Dis 1978; 37: 146-51. 1153-8. S Ebringer R, Cawdell D, Ebringer A. Klebsiella pneu- 13 Curvall M, Lindberg B, Lonngren J, Nimmich W. moniae and acute anterior uveitis in ankylosing spondy- Structural studies of the capsular polysacchararide of litis. Br Med J 1979; i: 383. klebsiella type 28. Carbohydrate Res 1975; 42: 95-105. 6 Eastmond CJ, Willshaw HE, Burgess SEP, Shinebaum R, 14 Buddecke E. Glycoproteins in mammals. In: Gottschalk Cooke EM, Wright V. Frequency of faecal Klebsiella A, ed. Glycoproteins. Amsterdam: Elsevier, 1972. aerogenes in patients with ankylosing spondylitis and controls with respect to individual features of the disease. 15 Hirata AA, McIntyre FC, Terasaki PI, Mittal KK. Cross- Ann Rheum Dis 1980; 39: 118-23. reactions between human transplantation antigen and 7 Ebringer A, Cowling P, Ngwa Suh N, James DCO, bacterial lipopolysaccharides. Transplantation 1973; 15: Ebringer RW. Crossreactivity between Klebsiella aero- 441-5. genes and B27 lymphocyte antigens as an aetiological 16 Boyden SV. Natural antibodies and the immune response. factor in ankylosing spondylitis. In: Dausset J, Svejgaard Adv Immunol 1966; 5: 1-28. A, eds. HLA and Disease. Paris: INSERM, 1976; 58: 27. 17 Rahi AHS, Garner A. Immunopathology of the Eye. 8 Welsh J, Avakian H, Cowling P, et al. Ankylosing London: Blackwell, 1976. spondylitis, HLA-B27 and klebsiella. I. Crossreactivity 18 Pearson CM, Wood FD. Studies of polyarthritis and studies with rabbit antisera. Br J Ex Pathol 1980; 61: other lesions induced in rats by injection of mycobacterial 85-91. adjuvant. Arthritis Rheum 1974; 17: 409-16. 9 Avakian H, Welsh J, Ebringer A, Entwistle CC. Anky- 19 Waksman BH, Bullington SJ. Studies of arthritis and losing spondylitis, HLA-B27 and klebsiella. II. Cross- other lesions induced in rats by injection of mycobac- reactivity studies with human tissue typing sera. Br J terial adjuvant. Arch Ophthalmol 1960; 64: 751-62. Exp Pathol 1980; 61: 92-6. 20 Rosenbaum JT, McDevitt HO, Guss RB, Egberg PR. 10 Seager K, Bashir HV, Geczy AF, Edmonds J, De Vere- Endotoxin induced uveitis in rats as a model for human Tyndall A. Evidence for a specific B27 associated cell disease. Nature 1980; 286: 611-3. http://bjo.bmj.com/ on September 24, 2021 by guest. Protected copyright.